dzynin / analysis-of-political-sentiments-on-twitter-using-deep-learning Goto Github PK
View Code? Open in Web Editor NEWThis project forked from bilalaziz786/analysis-of-political-sentiments-on-twitter-using-deep-learning
Aim was to develop a machine learning model which can analyze sentiments on twitter and to predict the winner of Lok Sabha Elections 2019. Web scraping was used for comments and then applied feature extraction, TF-IDF, Word2Vec, ANN and LSTM to improve the accuracy of model. Softmax function was used for multiclass classification. Learning curve was used to select the epochs and batch size in ANN and LSTM. Word cloud was made to select the most important features in the dataset. Random forest classification was selected as the best classifier with the help of PR curve by using micro average for average precision and recall.